Information Technology Reference
In-Depth Information
Remember the old saying: Half of your advertising is wasted—the trouble is you
do not know which half it is. the idea behind segmentation, especially online, is that
you might be able to figure out which half that is.
Any report you look at, whether out of the box or a custom report you created
and bookmarked, is likely to tell you completely different stories about your visitors
when looking at it from a segmented view. this segmented view can be as simple as the
day of week, the country of origin, or first-time or returning visits. generally an oppor-
tunity to split or combine visitors into groups is based on attributes like the following:
B
•
ehavior
d
•
emographics
•
p
sychographics
g
•
eography
m
•
oney
189
Segmentation Pointers
In case you are yet to pass your segmentation exam, here are a few pointers on how to get
started. Again, this is by no means everything there is to say about segmentation:
Use your existing objectives and goals as thresholds for how to approach segmentation. Say
•
that you have an overall site goal of a 1 percent visit-to-sale conversion rate and that your
site average for the given period is indeed a positive 1 percent. By the definition of the word
average, we know with certainty that some groups of visitors are converting above this ratio
and some are converting below this ratio. It is your task to isolate the better-performing
segments to learn from their attributes. It is also your task to find less optimal segments,
so you can eliminate the campaigns that drive these visitors or try to optimize the way you
serve them.
Define and create segments based partly on your intuition and understanding of the busi-
•
ness and partly by looking through sorted reports to spot winning sources, campaigns, and
other attributes.
Measure segmented behavior to the extent where you can conclude that you have a unique
•
segment that, for example, converts within a certain percentage range. This sounds easy,
but you should be very careful in how you do this—don't jump to conclusions based on too
little data or polluted data.
Create an optimization hypothesis on the poorer-performing segments, or just aggressively
•
cut support for campaigns and general activities that support this traffic influx.
Take action and execute the suggested optimization, and measure if the results live up to
•
the hypothesis using your newly defined segments.
